Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
qqr is an RL training framework for open-ended agents.
| Date | Stars |
|---|---|
| 2026-07-31 | 269 |
| 2026-08-06 | 269 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
15.0
growth rate 0.00%/day
# qqr
<h4 align="center">
<p>
<b>English</b> |
<a href="README_zh.md">中文</a>
</p>
</h4>
<p align="center">
<img src="assets/Logo.png" width="540"/>
<p>
<p align="center">
🤗 <a href="https://huggingface.co/collections/Alibaba-NLP/arenarl">HuggingFace</a> |
🤖 <a href="https://modelscope.cn/datasets/iic/Open-Travel">ModelScope</a> |
📰 <a href="https://tongyi-agent.github.io/blog/arenarl/">Blog</a> |
📑 <a href="https://huggingface.co/papers/2601.06487">Paper</a>
<p>
`qqr` (a.k.a. hilichurl) is a lightweight, non-intrusive extension for [`slime`](https://github.com/THUDM/slime). It seamlessly integrates the [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol) standard to enable the evolution of open-ended agents via [**ArenaRL**](https://arxiv.org/abs/2601.06487).
## 📰 News
- **[2026.07.30]** 🔥 We release [**SecRespond**](https://arxiv.org/abs/2607.26791), a benchmark for real-world post-compromise incident response, together with its [dataset and evaluation scripts](data/secrespond/).
- **[2026.05.01]** 🎉 Our paper ArenaRL has been accepted by **ICML 2026**!
## 🌟 Key Features
- **ArenaRL Algorithm**: Full implementation of the core algorithms described in the paper. It includes built-in topologies for Anchor-Based, Round-Robin, Swiss-System, Double-Elimination, and Seeded Single-Elimination tournaments.
- **Built for Open-Ended Agents**: Specifically engineered to tackle discriminative collapse in complex, open-ended tasks, ensuring continuous policy improvement via relative ranking even when reward model scores stagnate.
- **MCP Support**: Seamlessly integration with the [MCP]((https://github.com/modelcontextprotocol)) standardizes the decoupling of LLM inference and tool environments. Developers can reuse existing MCP Servers as training environments without rewriting interfaces.
- **High-Performance Training**: Built on top of [`slime`](https://github.com/THUDM/slime) to deliver high-throughput, distributed rollout generation and training for large-scale agent evolution.
## 📦 Installation
To get started, first ensure [`slime`](https://github.com/THUDM/slime) is installed (refer to [Quick Start](https://thudm.github.io/slime/get_started/quick_start.html)). Then install `qqr` from source:
```bash
git clone https://github.com/Alibaba-NLP/qqr.git
cd qqr
pip install -e .
```
## 🚀 Quick Start
Run the travel experiment quickly with the following command:
```bash
bash scripts/travel/run-qwen3-8B.sh
```
You can configure the experiment in [`qqr/examples/travel/config.py`](qqr/examples/travel/config.py).
## 📋 Compatibility
Due to `slime` version upgrades, specifically regarding rollout changes, please use the following version combinations that we have tested to ensure stability:
| qqr | slime |
| :----- | :----- |
| v0.2.0 | v0.3.0 |
| v0.1.3 | v0.2.4 |
| v0.1.2 | v0.2.3 |
| v0.1.1 | v0.2.2 |
| v0.1.0 | v0.2.1 |
## Acknowledgements
[**slime**](https://github.com/THUDM/slime): For providing a powerful post-training framework.
[**openai-agents-python**](https://github.com/openai/openai-agents-python): For providing excellent MCP interfaces.
## Citation
If you use `qqr` or the ArenaRL algorithm in your research, please cite our paper:
```bibtex
@misc{zhang2026arenarlscalingrlopenended,
title={ArenaRL: Scaling RL for Open-Ended Agents via Tournament-based Relative Ranking},
author={Qiang Zhang and Boli Chen and Fanrui Zhang and Ruixue Ding and Shihang Wang and Qiuchen Wang and Yinfeng Huang and Haonan Zhang and Rongxiang Zhu and Pengyong Wang and Ailin Ren and Xin Li and Pengjun Xie and Jiawei Liu and Ning Guo and Jingren Zhou and Zheng-Jun Zha},
year={2026},
eprint={2601.06487},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2601.06487},
}
```
Excerpt of 3,922 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:2481e5f6000ce5b8, llm:description: 'qqr is an RL training framework for open-ended agents.'
matched fp:2481e5f6000ce5b8, llm:description: 'qqr is an RL training framework for open-ended agents.'
matched fp:2481e5f6000ce5b8, llm:description: 'qqr is an RL training framework for open-ended agents.'